At present, the mode of transmitting the image data is mainly as follows: the image processing system receives a whole image data frame from the video camera and buffers it, and then transmits this image data frame to a server after processing and compressing it.
This mode of transmitting the image data has a longer delay. Take as an example that an image data frame at the resolution of 1024*768 is transmitted by the gigabit network (i.e., 1000 Mbps), where each pixel datum occupies 24 bits, the size of one image data frame is 1024*768*24 bits, and the delay consumed by transmitting this image data frame is 1024*768*24/1000, which is about 18 ms. That is, the transmission delay of one image data frame is 18 ms, which will be a fatal problem for the application scenario with a higher requirement for the real-time performance of image data transmission or for the application scenario where the image processing algorithm is time-consuming, such as data collection, automatic drive or another application scenario.